Katla Is Live on Product Hunt
We built a cookie consent platform that does not cost you Core Web Vitals. 4.9 kB headless, one request, and zero milliseconds added to LCP in our control-page tests. Here is what shipped, and what it measures against.
Katla is live on Product Hunt today: Cookie compliance without the complexity.
If you have ever added a consent banner to a fast site and watched the Lighthouse score fall over, you already know the problem we built this for. Consent is a legal requirement. Losing a second and a half of Largest Contentful Paint to it is not.
The number the whole thing is built around
We publish a benchmark rather than an adjective. Every platform below was loaded into one identical page, with the same content and the same GA4 and Meta pixel tag set, on a throttled mobile profile with a cold cache. The consent script is the only variable. Median of five runs, measured 4 September 2026.
| Platform | Transfer | Requests | Main thread |
|---|---|---|---|
| Katla, headless | 4.9 kB | 1 | not separately attributable |
| Katla, widget | 18.4 kB | 1 | 55 ms |
| CookieYes | 27.4 kB | 8 | 63 ms |
| Pandectes | 53.1 kB | 2 | 28 ms |
| Cookiebot | 151.4 kB | 4 | 391 ms |
| OneTrust | 171.6 kB | 7 | 125 ms |
Weight is the easy half. The half that shows up in field data is what a banner does to the page it sits on:
| Platform | LCP | LCP added | TBT | TBT added |
|---|---|---|---|---|
| Katla, widget | 299 ms | 0 ms | 165 ms | 17 ms |
| CookieYes | 299 ms | 0 ms | 242 ms | 94 ms |
| Pandectes | 694 ms | 0 ms | 255 ms | 0 ms |
| Cookiebot | 2,583 ms | 2,284 ms | 213 ms | 65 ms |
| OneTrust | 2,945 ms | 2,646 ms | 228 ms | 80 ms |
Two and a half seconds of added LCP is not a rounding error. It is the difference between a page that passes Core Web Vitals and one that does not, on a site whose developers did nothing wrong.
The full method, including the control page we built to make these comparable, is on the benchmark page. We re-run it, and we publish the date it was last measured, so you can tell how stale the numbers are.
What actually shipped
A headless SDK from 4.9 kB. You render your own banner with your own components. Katla handles the consent state, the storage, the signals and the blocking. There is a prebuilt widget too, at 18.4 kB, for when you would rather not draw one.
AI cookie scanning. Katla crawls your site, finds every cookie, and classifies each one with a purpose, a vendor and a retention period. Unknown cookies go to a model rather than into an "uncategorised" bucket nobody ever empties.
Policies generated from what you actually set. Your cookie policy is built from the scan, not from a template you edited once in 2023. When a developer adds a tag, the policy follows.
Google Consent Mode v2, and the checking to go with it. Getting the signal order wrong is easy and silent. There is a free checker for it.
GDPR, CCPA, DMA, and the APAC regimes. One configuration, the right behaviour per region.
The free things, no account needed
We would rather you check your own site before you believe any of the above:
- Cookie checker shows what fires before your banner is answered
- Consent Mode checker watches the gtag calls in order
- Compliance checker is an open source audit your coding agent runs locally
- Enforcement tracker is a running log of what European regulators have actually done about cookies
That last one is new this week, and it is the one we find ourselves reading. Most cookie enforcement is not a headline fine. It is a reprimand, an order to fix something, or a case closed quietly once the company changed its banner. Sweden's telecom regulator closed four such cases in a single day in 2023, with no penalty at any of them, over things like a reject button that was not in the first view.
Come and argue with us
The benchmark is the part we most want challenged. If your numbers differ from ours, we want to know, because a benchmark nobody can reproduce is marketing with a table around it.
We are on Product Hunt today. An upvote helps, a comment helps more, and finding a hole in the methodology helps most of all.